TBNAI
CANOPY: Adaptive-Granularity Evidence Compression for Multimodal RAG
Publication: Hugging Face Daily Papers
Author: Not available in collected metadata
Source published: 30/09/2026, 20:00:00 UTC
Briefing created: 17/10/2026, 01:46:41 UTC
Based on a feed excerpt; full article content has not been acquired. Source evidence retrieved 06/10/2026, 04:18:26 UTC.
Briefings
CANOPY is a framework for adaptive-granularity evidence compression in multimodal RAG systems. It represents retrieved items as hierarchies and uses a node encoder fine-tuned on gold evidence to score regions against the query. Parent-relative refinement selects multiple regions at different granularities without LLM calls for node-level pruning. Across five QA benchmarks over a 33M-item heterogeneous corpus, CANOPY achieves higher average answer accuracy than evaluated retrieval baselines. Ablations indicate that additional retrieval drives the main accuracy gains on multi-hop QA. In the unrouted Qwen3-VL-8B-Instruct setting, compression reduces reader-input evidence tokens by 14.2-27.7% relative to the same iterative pipeline without compression, with comparable answer accuracy.
Key points
- CANOPY uses a node encoder fine-tuned on gold evidence to score regions against the query for adaptive-granularity compression.
- Parent-relative refinement selects multiple regions at different granularities without LLM calls for node-level pruning.
Context
Multimodal RAG retrieves text, tables, images, and videos, but choosing a retrieval granularity does not determine how much context to retain within each item. Existing compressors address this trade-off with modality-specific mechanisms, leaving open a shared procedure for adapting the retained extent region by region across heterogeneous items.
Limitations
Compression cannot recover evidence that was never retrieved, so a critic requests targeted follow-up retrieval when it judges the accumulated evidence insufficient. Results are from five QA benchmarks over a 33M-item heterogeneous corpus, and the 14.2-27.7% reduction applies specifically to the unrouted Qwen3-VL-8B-Instruct setting.
Generated by Bittrees-hosted qwen35-2b sha256:aaf42c8b7c3c on 17/10/2026, 01:46:41 UTC. Based on publisher-provided feed evidence; may not cover the full original article. Check the source for details.
View archived briefing on IPFS
bafybeihscvtgxxpc636vticife24fft65dunhraponyo6g2k54jyxu23rm